Enterprises have backtracked on their diversity commitments over the last couple of years. Economic and political pressure drove it. The performance data on diverse teams is the same as it was and the companies who are still committed have stopped using the phrase “Diversity & Inclusion” and are now using softer language.
For engineering, this was always a commercial question. Diverse teams build better software and make better decisions under pressure than teams where everyone in the room thinks the same way. The external politics have changed but the business case for diversity hasn’t. Some people have just stopped caring about team composition because they no longer feel they “have to”.
When researchers ran experimental stock markets across two continents, prices in ethnically diverse markets tracked true value 58 per cent more accurately than in homogeneous ones, because traders in homogeneous rooms trusted each other’s calls too readily and mispriced together. When everyone reasons the same way, mistakes correlate and get expensive and it’s the same dynamic on an engineering team.
Performative DEI vs hiring for performance
None of that tells you whether diversity on your own team matters. Whether a formal DEI programme was the right vehicle is one question. Whether the make-up of your engineering team affects how well it performs is a different one.
There’s a huge difference between performative DEI and diverse hires made for performance. One is a PR initiative about branding. The other uses the makeup of your engineering team to improve outcomes and reduce costs, because different backgrounds catch different things. I still work with plenty of businesses serious about building diverse teams. They’ve just stopped calling it DEI, because it carries political and commercial risk.
What changes when your team is diverse
The teams I’ve seen build well over the years aren’t the ones with the strongest individuals. They’re the ones made up of different backgrounds, and the team is set up to draw on those differences.
In practice, that’s an employee on the autism spectrum catching logical inconsistencies before a single line of code is written, because pattern recognition is how they think. It’s an engineer catching the kind of algorithmic blind spot that saw early facial-recognition systems misclassify darker-skinned women up to a third of the time, while reading lighter-skinned men almost perfectly. It’s a Korean developer whose own name doesn’t fit a ‘first name, last name’ form catching the data model before it shipped and locked out millions of users worldwide. It’s a woman on a male-dominated engineering team pushing back on defaults in systems trained mostly on male data.
Caroline Criado Perez’s Invisible Women documents how products built around a male default fail everyone else. Voice recognition trained on men’s voices that struggles with women’s. Car safety designed around a male-sized crash dummy. Those are exactly the failures a room full of similar people never thinks to check for.
When everyone in the room is a thirty-something male computer science graduate from the same companies, their blind spots are all going to be very similar.
A strong new hire changes how a team operates
A single hire from outside the existing culture changes how the whole team behaves. If you want to change how a team operates, change who’s in it.
Put a senior engineer with regulated-industry experience into a team that’s only ever built consumer products, and the standup conversation changes. Decisions get challenged in ways they weren’t before, and the kind of candidate you can attract next changes too. Team behaviour is shaped by the people, which is why training and process never move it the way a good hire can.
How you hire for it
Start by asking your team a direct question: which traits, skills and characteristics do we all share, and what might we be missing because of it? It’s not a comfortable question, which is why it doesn’t get asked. It’s a far more useful starting point than a generic diversity policy, because it links the conversation to the actual work. Some teams use Gallup’s CliftonStrengths to map where they overlap and where the gaps are.
Referrals are great, but they tend to reproduce the team you already have. People refer people they know, who are usually like themselves. You need multiple targeted channels to find people you wouldn’t otherwise attract.